I-DWRL: Improved Dual Wireless Radio Localization Using Magnetometer

Transcription

1 sesors Article I-DWL: Improved Dual Wireless adio Localizatio Usig Magetometer Abdul Aziz, amesh Kumar ad Iwhee Joe 3, * Departmet Electroics ad Computer Egieerig, Hayag Uiversity, Seoul 04763, Korea; azizsheraz@yahoo.com Departmet Electroic Egieerig, Dawood Uiversity Egieerig ad Techology, Karachi 74800, Pakta; rameshkarmai@gmail.com 3 Divio Computer Sciece ad Egieerig, Hayag Uiversity, Seoul 04763, Korea * Correspodece: iwjoe@hayag.ac.kr; Tel.: eceived: 30 August 07; Accepted: 3 November 07; Publhed: 5 November 07 Abstract: I dual wireless radio localizatio (DWL techique each sesor ode equipped with two ultra-wide bad (UWB radios; dtace betwee two radios a few tes cetimeters. For localizatio, DWL techique must use at least two pre-localized odes to fully localize a ulocalized ode. Moreover, i DWL techique it also ot possible for two sesor odes to properly commuicate locatio iformatio uless each four UWB radios two commuicatig sesor odes caot approach remaiig three radios. I th paper, we propose a improved DWL (I-DWL alog with moutig a magetometer sesor o oe UWB radios all sesor odes. Th additio a magetometer helps to improve DWL such that oly oe localized sesor ode required for localizatio a ulocalized sesor ode, ad localizatio ca also be achieved eve whe some four radios two odes are uable to commuicate with remaiig three radios. The results show that with use a magetometer a greater umber odes ca be localized with a smaller trasmsio rage, less eergy ad a shorter period time. I comparo with covetioal DWL, our I-DWL ot oly maita localizatio error but also requires aroud half semi-localizatios, 60% time, 70% eergy ad a shorter commuicatio rage to fully localize a etire etwork. Moreover, I-DWL ca eve localize more odes while trasmsio rage ot sufficiet for DWL. Keywords: DWL; I-DWL; UWB; dual radio; magetometer; localizatio; wireless sesor etwork. Itroductio Wireless sesor etworks have bee a vital research area for may years. Nowadays sesor odes are used widely i variety applicatios, devices, buildigs, vehicles, ad so forth. The first research i th area was motivated by military applicatios, with defese advaced research projects agecy (DAPA fudig a umber promiet research projects such as Smart Dust, ad eural egieerig sciece ad techology (NEST []. Wireless sesor etworks (WSNs cost collectio small, low-cost, usually radomly placed heterogeeous sesor odes coected by wireless media to form a sesor field. Nodes moitor eviromet, gar data ad sed it to sik ode through sigle or multichip commuicatios. Its may applicatios iclude battlefield surveillace, habitat moitorig, evirometal moitorig, health applicatios, target trackig, evet detectio, vehicle trackig, ad forest fire detectio []. To make collected iformatio valuable, may applicatios such as geographical routig ad locatio-based iformatio applicatios require eact locatios deployed sesor odes. Therefore, may locatio-fidig schemes have bee proposed for wireless sesor etworks. Sesors 07, 7, 630; doi:0.3390/s7630

2 Sesors 07, 7, Th paper orgaized as follows. Sectio covers related work ad details dual wireless radio localizatio (DWL. A improved dual wireless radio localizatio (I-DWL scheme preseted i Sectio 3. Sectio 4 presets simulatio work ad results, ad we coclude paper i Sectio 5.. elated Work Localizatio i WSN a active area research with several surveys [3 ] ad more tha 50 localizatio s. Localizatio techiques ca be geerally categorized as followig: absolute vs. relative cetralized vs. decetralized rage free vs. rage based achor vs. achorless. Absolute localizatio to locate sesor odes with respect to coordiate system, whereas i relative localizatio, locatio sesor odes foud i relatio to or sesors. Global positioig system a eample absolute localizatio []. I cetralized localizatio [3,4], oe cetral base statio preset for computatio. The dadvatage overhead, ad cost also icreases. Multidimesioal scalig map (MDS-MAP [4] a cetralized for computig coordiates ukow odes after approimatig dtaces betwee odes usig shortest path. I cotrast, decetralized or dtributed localizatio techiques [5,6] deped o each sesor ode beig able to determie its locatio with oly limited commuicatio with earby odes. age-free [7] techiques deped upo factors such as umber hop couts ad coectivity, whereas rage-based techiques iclude received sigal stregth idicatio (SSI [8], time arrival (ToA [9], agle arrival (AoA [0], time differece arrival (TDoA, lateratio ad agulatio, ad so forth [0]. There wide rage radio sigals but ultrawide bad (UWB sigals are particularly well suited for rage-based localizatio, sice y ca provide accurate ad reliable rage measuremets due to ir fie delay resolutio ad robustess i harsh eviromets [ 4]. The achor-based s provide a startig poit for a by usig positio achor odes. I cotrast, achorless schemes measure dtace betwee odes for creatig a local map odes. The local map created ot a uique oe ad ca be stitched to ay coordiate system with help traslatio, rotatio or flippig. DWL ad I-DWL s fall ito category rage-based localizatio where a achor ode also used to start localizatio process. Both s are desiged for coplaar static wireless sesor ad ad hoc etworks... Dual Wireless adio Localizatio The dual wireless radio localizatio (DWL [5] a Global Positioig System (GPS free-rage based dual radio wireless localizatio for static wireless etworks, where each ode has two UWB wireless radios amed adio ( ad adio ( that are attached to a a priori kow positios o board, as show i Figure. The straight lie joiig ad cosidered as a ode. I th techique oe odes desigated as sik ode whose set at origi (, y = (0, 0 ad i directio positive -a (, y = (d, 0, which meas agle betwee -a ad a ode zero ad, here, d dtace betwee ad. Two odes are said to be colliear if re a straight lie passig through all four radios odes. The authors DWL have eperimetally show i ir paper ad suggested that, for a typical wireless sesor ode, iter-radio dtace 60 cm sufficiet for successful localizatio [5]. While a icreased iter-radio dtace improves localizatio accuracy, i all our

3 Sesors 07, 7, Sesors 07, 7, 630 ( =(0,0 d ( =(d,0 3 5 simulatio scearios we cosider Figure miimum. Dual iter-radio wireless radio dtace ode. 60 cm to avoid a greater size sesor odes for practical scearios. DWL Two odes are said Sesors 07, 7, 630 preseted to be colliear i three if re steps, a which straight are lie preseted passig below. through all four radios 3 6 odes. The authors DWL have eperimetally show i ir paper ad suggested that, for a typical wireless sesor ode, iter-radio dtace 60 cm sufficiet for successful localizatio [5]. While a icreased iter-radio dtace improves localizatio accuracy, i all our ( d =(0,0 ( =(d,0 simulatio scearios we cosider miimum iter-radio dtace 60 cm to avoid a greater size =0 sesor odes for practical scearios. DWL preseted Figure i three. Dual steps, wireless which radio are preseted ode. below. Figure. Dual wireless radio ode.... Two Semi-Localizatio odes are said to be colliear if re a straight lie passig through all four radios... Semi-Localizatio odes. The authors DWL have eperimetally show i ir paper ad suggested The first step towards localizatio ay ulocalized ode with help a sik ode or that, The for a first typical stepwireless towardssesor localizatio ode, iter-radio ay ulocalized dtace ode 60 with cm sufficiet help for sik successful ode or some or already localized ode semi-localizatio. For localizatio oly, UWB radio ragig some localizatio or [5]. already While localized a icreased ode iter-radio semi-localizatio. dtace improves For localizatio localizatio oly, accuracy, UWB radio i all ragig our used to measure four dtaces r simulatio used to measure scearios four we dtaces r, cosider miimum, r r,, r r ad 3 r 4 betwee radios two odes that are 3 aditer-radio 4 betwee dtace radios 60 cm to avoid two odes a greater thatsize are withi withi sesor rage rage odes each each or, for practical or, as as show scearios. show i i Figure Figure.. Here Here sik sik ode, ode, ad ad a a ulocalized ulocalized ode, DWL where d ad ad d d are preseted are already already i kow kow three steps, radio radio which separatio are preseted dtaces. below. By By usig law cosies as give i Equatios ( (8, two solutios are obtaied for locatio ode. Figure shows... semi-localizatio Semi-Localizatio betwee ode ad. ad. The first step towards localizatio ay ulocalized ode with help a sik ode or some or already localized ode semi-localizatio. For localizatio oly, UWB radio ragig used to measure four dtaces r, r, r ad 3 r 4 betwee radios two odes that are withi rage each or, as show i Figure. Here sik ode, ad a ulocalized ode, where d ad d are already kow radio separatio r 3 dtaces. By usig law cosies as give i Equatios ( (8, two solutios are obtaied for r locatio ode. Figure shows semi-localizatio betwee ode ad. =0 d r 4 = d +r dr cos d =0 d +r3 r d3 r d ( ( r rr = d +r3 dr3 cos r 3 r = d +r 3 d r 3 cos ( ( r 4 = d +r d r cos ( ( ( = = ± cos ( d +r 3 r (3 (3 Figure. Semi Localizatio betwee d r 3 ode ad. = ± cos ( d d +r +r r (4 4 r = d cos +r ( 3 d d r3cos r d ( (4 r = r cos (5 r 4 = d +r dr cos ( y = r cos (5 = ±r si (6 d +r3 r = cos y = r ( 3 cos (7 (3 = r si d 3 r (6 y = ±r 3 si (8 Figure 3 shows two solutios for positio d ode +r r 4 = cos ( : oe actual positio ad or (4 positio located symmetrically o a flip aroud a d r ode. There will be oly oe solutio if ad oly if aes ad lie o a same straight lie or, i or words, both odes are colliear. = r cos (5 ( ( =0 d ( ( r Figure. Semi Localizatio betwee ode ad. Figure. Semi Localizatio betwee ode ad. ( y = r si (6

4 y = r si (8 3 Figure 3 shows two solutios for positio ode : oe actual positio ad or positio located symmetrically o a flip aroud a ode. There will be oly oe solutio Sesors if ad 07, oly 7, 630 if aes ad lie o a same straight lie or, i or words, both odes 4are 5 colliear. ( d r 3 ( r r d ( ( =0 r (, y r 3 r d (, y Figure 3. Two possible positios ode (i.e., ad. Figure 3. Two possible positios ode (i.e., ad.... igid Localizatio... igid Localizatio As i semi-localizatio, if two odes ad are colliear, both locatio As i semi-localizatio, if two odes ad are colliear, both locatio solutios solutios will be same, ad will idicate actual positio ode. If both odes are ot colliear will be same, ad will idicate actual positio ode. If both odes are ot colliear re re a eed rigid localizatio. igid localizatio makes use a additioal ode with a eed rigid localizatio. igid localizatio makes use a additioal ode with additioal additioal semi-localizatio for selectig correct solutio for ode by matchig locatio semi-localizatio for selectig correct solutio for ode by matchig locatio that best that best overlapped from ad 3 semi-localizatio steps, where odes overlapped from ad 3 semi-localizatio steps, where odes ad 3 are already localized. ad 3 I are Figure already 4 ode localized. I Figure 4 ode rigid localized with help ode ad rigid localized with help ode ad 3, which are somehow 3 already, which rigid are localized. somehow Here, already rigid localized. Here, ad are ot actual positios ad are ot actual positios.. Figure 4. igid localizatio ode usig localized odes ad DWL Algorithm The sik ode iitializes localizatio process. Firstly, oe eighbor odes semi-localized ad declared to be rigid localized eve without performig rigid localizatio, because sik

5 Figure 4. igid localizatio ode usig localized odes ad DWL Algorithm Sesors 07, 7, The sik ode iitializes localizatio process. Firstly, oe eighbor odes semilocalized ad declared to be rigid localized eve without performig rigid localizatio, because oly sik localized oly ode localized ad re ode ad o or re localized o or ode localized to perform ode to rigid perform localizatio. rigid localizatio. The iitialthe guess iitial actual guess positio actual positio first ode first leftode to be verified left to be byverified a thirdby party a third outside party outside etwork such etwork as etwork such as operator, etwork ad operator, it should ad be it should verified be at verified ed at whe ed whe whole etwork whole etwork localized. localized. Later, sik ad first localized ode will rigid localize ir eighbor odes ad Later, sik ad first localized ode will rigid localize ir eighbor odes ad process process cotiues. Durig localizatio process, all ulocalized odes keep lteig uless y cotiues. Durig localizatio process, all ulocalized odes keep lteig uless y receive receive message from two localized eighborig odes. Fially, a third party should check wher message from two localized eighborig odes. Fially, a third party should check wher iitial iitial guess locatio first ode was ot correct. I th case, locatios all odes guess locatio first ode was ot correct. I th case, locatios all odes should be should be symmetrically flipped aroud a sik ode. Figure 5 shows flowchart DWL symmetrically flipped aroud a sik ode. Figure 5 shows flowchart DWL.. Start Sik ode iitializes localizatio process so that all reachable ulocalized odes should keep lteig uless y receive message from two localized eighbors. Sik semi localizes oe it s eighborig ode ad declares it as a rigid localized ode with iitial guess oe it s two possible locatios. Sik ad ode carryout furr rigid localizatio for eighborig odes ad process cotiues hop to hop till all odes are localized Was iitial guess locatio ode right? No Symmetrically flip locatios all odes aroud a sik ode Yes Ed Figure 5. Flowchart dual wireless radio localizatio (DWL. Figure 5. Flowchart dual wireless radio localizatio (DWL...4. DWL Algorithm Drawbacks Although dual radios o a sigle ode for localizatio effective, DWL has followig drawbacks: a. If iitial locatio sik ode ot kow, DWL cosiders that adio ( si k sik ode has bee assiged a specific locatio (0,0, ad adio ( si k sik ode cosidered i directio positive -a, which poits to local east directio. If odes are radomly deployed, how ca we suppose adio sik ode i directio local east? The wrog agle a sik ode ca lead to wrog locatio rest etwork odes. Therefore, we eed to fid actual directio si k. b. The DWL eeds at least two localized odes to fully localize ad ulocalized ode. For th reaso, DWL caot rigid localize first ode with oe sik ode. First, semi-localized ode declared as rigid localized o bas oe radomly chose locatio solutio two possible locatio solutios first ode. Th ca lead to serious localizatio problems if rest etwork cotiues to localize with help wrog locatio first ode. What if applicatios sceario critical, ad we have to use locatio some odes before whole etwork localizatio process completed, ad third party fids out that locatio first ode was ot right ad locatio all odes eed to be flipped aroud a sik ode? Therefore, we eed to fid eact locatio first ode.

6 Sesors 07, 7, c. To rigid localize a ode, a miimum two semi-localizatios are required. If a ulocalized ode caot lte from at least two localized odes it caot be localized ad it has to wait, uless somehow two localized sigals are received. If some ode caot receive two localized sigals it caot be localized at all. Th requires DWL to perform at least ( 3 semi-localizatios [5] to localize etire etwork. Here, ( 3 umber semi-localizatios required for etwork with umber odes, which equal to twice umber odes ad three less, where three idicates that re are o semi-localizatios performed for a sik ode ad oly oe semi-localizatio performed for ode that will be localized first. Sice ( 3 semi-localizatios ca be carried out with each choice a differet edge, total umber semi-localizatios ca be, at most, P ( 3. A higher umber semi-localizatios uses more eergy ad time, refore we eed to miimize total umber semi-localizatios as well as icrease umber fully localized odes eve with sigle semi-localizatio step. d. For successful semi-localizatio, each four radios two odes must be i commuicatio rage with rest three radios. Th requires eir high ode desity or high trasmsio power, where high ode desity costs more odes ad high trasmsio power reduces life sesor odes. We eed to develop a way for odes to be localized eve if few four radios two commuicatig odes ca reach each or. 3. Improved DWL (I-DWL Algorithm To preset I-DWL, we ited to improve drawbacks DWL, as metioed earlier. To achieve such improvemets, we use a magetometer, which affied o adio every sigle ode, as show i Figure 6. c agle measured with magetometer ad ( = 360 c agle which roughly idicates slope a ode; it will be used i localizatio process. Sesors 07, 7, sik sik sik sik d ( =( c 0 70 W 0 N S 80 =0 E 90 ( =( (a d =30 sik sik ( =( 70 W 330 N 0 c sik sik ( =( (b S 80 E 90 Figure 6. Node with Magetometer showig differet agles magetometer c ad Figure 6. Node ode i (a,b. with Magetometer showig differet agles magetometer c ad ode i (a,b. I I-DWL, which dtict from origial DWL, we have added iitial locatio adio sik sik sik I I-DWL, sik ode which ( as dtict ( from, which origial fed DWL, eact we have locatio added usig GPS iitial or ay locatio or meas adio sik locatio sik ode source, ( si ad k as ( si iitial k, y si locatio k, which adio fed eact sik ode locatio ( usig measured GPS or ay usig or simple meas locatio trigoometry, source, as adgive iitial Equatio locatio (9. adio Furr sik Equatios ode ( si (5 (8 k measured are modified usig simple ito trigoometry, Equatios (0 (3 as giveto iaccommodate Equatio (9. ay Furr iitial Equatios positio (5 (8 ad agle are modified a ito sik Equatios ode. The (0 (3 iitial to accommodate positio a sik ayode iitialca positio help to ad locate agle or a odes sik with ode. ir The real iitial positios positio i coordiate a sik ode casystem. help toaddig locate or actual odes locatio with ir a sik realode positios solves i first coordiate drawback system. DWL Addig. actual locatio a sik ode solves first sik sik drawback sik DWL sik. ( = ( +d cos +d si (9 = r cos( + + (0a = r cos( + + (0b

7 Sesors 07, 7, ( si k, y si k = ( si k +d cos, y si k +d si (9 = r cos( + + (0a = r cos( + + y = r si( + + y y = r si( + + y = r 3 cos( + + = r 3 cos( + + y = r 3 si( + + y (0b (a (b (a (b (3a y = r 3 si( + + y A magetometer cheap but susceptible to evirometal oe, ad it may require calibratio from time to time. To deal with problem magetometer calibratio, we cosider that all magetometers are calibrated well before localizatio process starts, ad magetometers should be calibrated wheever required ad feasible. To deal with problem agle errors due to evirometal oe, we cosider that we have kowledge evirometal oe ad we kow effect Sesors such 07, 7, oe 630 o magetometer agle deflectio. With th kowledge, it appears that 8 we 6 ca calculate, i advace, maimum possible magetometer agle errors due to evirometal oe. If a at a agle obtaied from magetometer at ± ad d_ma maimum possible agle agle deflectio caused due to to evirometal oe, we compare which slope two possible = positios ulocalized ode ± d_ma cosidered as a falls withi agle rage rage d_ma cosidered as a actual actual locatio locatio without performig rigid localizatio. Figure 7 depicts such a sceario, where without performig rigid localizatio. Figure 7 depicts such a sceario, where slope S slope ad S ad S ca be foud usig Equatio (4. S ca be foud usig Equatio (4. (3b =5 d S =5 rage d_ma 5 0 =0 d =5 Magetometer S =0 0 rage d_ma 5 0 d S = 5 Figure 7. The magetometer sesor helps to dtiguh betwee actual ad aor possible locatio Figure 7. The magetometer sesor helps to dtiguh betwee actual ad aor possible locatio ode. ode. y y S = ta ( (4a ( y ( y S = ta ( (4b

8 Sesors 07, 7, S = ta ( y y (4a S = ta ( ( y ( y (4b For such semi-localizatio to produce accurate results rigid localizatio without performig rigid localizatio, miimum differece betwee agle slope a localized ode ad agle magetometer a ulocalized ode must be greater tha d_ma. As with use a magetometer, we ca use oly oe semi-localizatio to rigid localize a ulocalized ode, refore, it solves secod ad third drawbacks DWL. To overcome coectivity problem DWL, we have cosidered si cases, four which are show i Figure 8. Sesors 07, 7, =45 ( r d ( d ( ( =5 r ( (a d =45 ( ( r d ( d ( ( =5 r ( (b d =5 ( =5 =5 ( ( r d d r =0 =0 ( ( d ( ( ( ( =5 r d d r ( d =80 ( ( =80 ( (c (d Figure 8. Various cases coectivity problem. (a Case a. Two radios localized ode ca reach Figure 8. Various cases coectivity problem. (a Case a. Two radios a localized ode ca reach oly adio ulocalized ode. (b Case b. Two radios localized ode ca reach oly adio oly adio ulocalized ode. (b Case b. Two radios a localized ode ca reach oly adio a ulocalized ode. (c Case c. adio localized ode ca reach oly adio a ulocalized a ulocalized ode. (c Case c. adio a localized ode ca reach oly adio a ulocalized ode, whereas adio localized ode ca reach both radios a ulocalized ode. (d Case d. ode, whereas adio a localized ode ca reach both radios a ulocalized ode. (d Case d. adio a localized ode ca oly reach adio a ulocalized ode, whereas adio a adio a localized ode ca oly reach adio a ulocalized ode, whereas adio a localized localized ode ca reach both radios a ulocalized ode. ode ca reach both radios a ulocalized ode. For Case a, actual ad or possible positio a ulocalized ode ca be foud with help a localized ode usig Equatios (4, (0 ad (, whereas actual ad or possible locatio a ulocalized ca be foud usig agle i Equatios (5 ad (6. = +d cos (5a

9 Sesors 07, 7, For Case a, actual ad or possible positio a ulocalized ode ca be foud with help a localized ode usig Equatios (4, (0 ad (, whereas actual ad or possible locatio a ulocalized ca be foud usig agle i Equatios (5 ad (6. = +d cos = +d cos y = y +d si (5a (5b (6a y = y +d si The decio regardig choosig actual locatio out two obtaied locatios made o bas that adio ad adio ode caot reach adio ode, but i obtaied locatio adio ode lies withi commuicatio rage adio ode. Mamatically, actual locatio ode ca be cirmed usig Equatio (7. I Equatio (7, if coditio true actual locatio else. (6b + > + (7 For Case b, actual ad or possible locatio ad ode ca be foud usig Equatios (4 ad (8 (. Moreover, we ca decide actual positio ode usig Equatio (. I Equatio ( if coditio true actual locatio else. = r cos( + + = r cos( + + y = r si( + + y y = r si( + + y (8a (8b (9a (9b = +d cos( 80 (0a = +d cos( 80 (0b y = y +d si( 80 (a y = y +d si( 80 (b + > + ( For Case c, actual ad aor possible locatio ode foud usig Equatios (4, (0, (, (5 ad (6, where Equatio (3 helps to decide actual positio ode. I Equatio (3 if coditio true actual locatio, orwe. + < + (3 For Case d, Equatios (4, ad (8 ( are used to fid actual ad or possible locatios, where Equatio (4 used to decide actual positio ode. I Equatio (4, if coditio true actual locatio else. + < + (4

10 Sesors 07, 7, Case e ad f ca also be cosidered with little variatio to Case c ad d, where adio a localized ode ca reach both radios a ulocalized ode, ad adio a localized ode ca reach eir adio or adio ulocalized ode, respectively. Same equatios will be used for Case e ad f, as i Case c ad d. All si se cases are used to solve coectivity sue DWL s i situatios where ode desity or trasmsio rage ot eough. A detailed flowchart DWL show i Figure 9. Sesors 07, 7, Start All odes keep lteig for localizatio messages uless y are localized. Sik / localized ode iitiates localizatio process by sedig its locatio iformatio ad maimum possible agle magetometer deflectio. d_ma No Does situatio match ay 06 cases that solve coectivity problem? Yes No eceivig ode checks if all four radios ca commuicate? Yes Ed Forward locatio message to localize furr odes Localize, a ulocalized ode lte for aor better localizatio message to reduce locatio error ad, th first localizatio message? Yes ad, agle betwee two odes No No igid localize Wait for or localizatio message > d_ma Ed? Yes Localized ode act like sik to carry o furr localizatio process Message receivig ode performs semi localizatio U localized ode ow localized Yes Slope Of ay two possible locatios falls with i? rage No Figure 9. Flow chart improved (I-DWL. Figure 9. Flow chart improved (I-DWL. 4. Simulatio Sceario, Parameters ad esults 4. Simulatio Sceario, Parameters ad esults I th sectio, we simulate DWL ad I-DWL s ad aalyze effects parameters I th such sectio, as rage we simulate measuremet DWL errors, ad magetometer I-DWL s errors, trasmsio ad aalyze rage, ad effects so o, where parameters NS- [6] such as used rage as a measuremet simulatio tool. errors, We compare magetometer DWL ad errors, I-DWL trasmsio i terms rage, umber ad so o, semi-localizatios where NS- [6] required, used as a simulatio percetage tool. etwork We compare odes DWL localized, ad ad I-DWL time i terms ad eergy required umber to semi-localizatios fih localizatio process. required, To accommodate percetage rage etwork measuremet odes localized, errors, ad real-world time oe ad eergy cosidered required as to fih summatio localizatio high process. probability To accommodate small oe ad rage low measuremet probability large errors, oe real-world [7,8]. Small oe modeled as a Gaussia radom process as a fuctio, which wireless rage odes, which give i Equatio (5. NS = 0.0 l (5

11 Sesors 07, 7, oe cosidered as summatio high probability small oe ad low probability large oe [7,8]. Small oe modeled as a Gaussia radom process as a fuctio, which wireless rage odes, which give i Equatio (5. N S ( = 0.0 l( (5 The low probability large oe value selected as a fuctio with probability less tha 0.05, which give i Equatio (6. Sesors 07, 7, 630 N L ( = 0.05 (6 6 Furr magetometer maimum possible deflectio from its accurate agle a joit effect both oes, ad it it modeled as as a a fuctio both both oes oes d_ma, N L. d_ma (N S,N L. The effect both oes cosidered i i such such a way way that that 00< d_ma 45. d_ma 45. Figure 0 ad Table shows simulatio sceario ad parameters, respectively. Figure 0 ad Table shows simulatio sceario ad parameters, respectively. Figure 0. Network simulatio sceario. Figure 0. Network simulatio sceario. Table. Network simulatio parameters. Ultra-wide bad (UWB. Table. Network simulatio parameters. Ultra-wide bad (UWB. Simulatios Parameter Value SimulatiosArea Parameter 50 Value 50 m Ulocalized Area odes m Ulocalized Localized odes/sik 3 Localized Node odes/sik dtributio Uiform radom dtributio Node dtributio Uiform radom dtributio Agle a odes radom Agle a odes radom Differet trasmsio rages rages 8, 8, 0, 0,.5,.5, 3.5, 3.5, 4.5, 4.5, m m Iter-radio dtace cm cm adio adio UWB UWB UWB age error without evirometal oe % trasmsio rage UWB age error without evirometal oe % trasmsio rage Figure shows that with lower trasmsio rage coectivity problems ca occur, ad a smaller umber odes will be localized with DWL, whereas I-DWL performs well eve with smaller trasmsio rage rage odes. odes. With With trasmsio rage rage 8 m, 8 m, DWL caot eve localize more tha 5% etwork odes, whereas our I-DWL ca localize aroud 80% etwork odes.

12 Sesors 07, 7, DWL caot eve localize more tha 5% etwork odes, whereas our I-DWL ca localize aroud 80% etwork odes. Sesors 07, 7, Sesors 07, 7, Figure.. Trasmsio rage versus versus total total percetage percetage localized localized odes. odes. Figure Trasmsio rage Figure. Trasmsio rage versus total percetage localized odes. As we have improved DWL i terms havig a smaller umber semias improved DWL i terms havig smaller aumber semi-localizatios Aswe wehave have improved DWL i terms ahavig smaller umber semilocalizatios required for full localizatio ulocalized odes, refore, Figure shows that required for full localizatio ulocalizedodes, refore, Figure shows that I-DWL localizatios required for fulllocalizatio ulocalized odes, refore, Figure shows that I-DWL ca localize higher percetage odes with a lower umber semi ca localize higher percetage odes with a lower umber eve I-DWL ca localize higher percetage odes with a semi-localizatios, lower umber semilocalizatios, eve i oy eviromets. I a ideal case where trasmsio rage 5 m ad i oy eviromets. I a ideal case where trasmsio 5 m adrage re 5 o m oe, localizatios, eve i oy eviromets. I a ideal case whererage trasmsio ad re o oe, I-DWL ca save aroud 50% semi-localizatios. I worst case, I-DWL ca save aroudca 50% aroud semi-localizatios. I worsti case, wherecase, re o oe, I-DWL save 50% semi-localizatios. worst where trasmsio rage 0 m, DWL ca reach oly 9% etwork odes; trasmsio rage 0rage m, DWL reach oly 9% oly etwork refore, where trasmsio 0 m, DWL ca ca reach 9% odes; etwork odes; refore, it performs a smaller umber semi-localizatios, while I-DWL ca reach 00% itrefore, performsita performs smaller umber semi-localizatios, while I-DWLwhile ca reach 00%ca etwork odes a smaller umber semi-localizatios, I-DWL reach 00% etwork odes ad performs a greater umber semi-localizatios. If agle deflectio ad performs a greater umbera semi-localizatios. If agle deflectio etwork odes ad performs greater umber semi-localizatios. If agle magetometer deflectio magetometer icreased, performace I-DWL degrades ad requires a icreased, performace I-DWL degrades ad requires a greater umber magetometer icreased, performace I-DWL degrades ad requires a greater umber semi-localizatios. semi-localizatios. greater umber semi-localizatios. Figure. Trasmsio rage versus umber semi-localizatios i oeless ad oy eviromet. Figure. Trasmsio rage versus umber semi-localizatios i oeless ad oy eviromet. Figure. Trasmsio rage versus umber semi-localizatios i oeless ad oy eviromet. For a trasmsio rage 5.5 m, Figure 3 shows that I-DWL requires less For a trasmsio rage 5.5 m, Figure 3 shows that I-DWL requires less tha For 60% etwork localizatio timem, ad takes 3 aroud 70% eergy to fully localize same trasmsio rage 5.5 shows that requires less tha 60%a etwork localizatio time adfigure takes aroud 70% I-DWL eergy to fully localize same etwork as with DWL. tha 60%as with etwork localizatio time ad takes aroud 70% eergy to fully localize same etwork DWL. For as with simulatio results localizatio error, we have cosidered two trasmsio rages, etwork DWL. For simulatio results localizatio error, we have cosidered two trasmsio rages, amely, 5.5simulatio m ad 4.5 m. The magetometer deflectio due to evirometal For agle localizatio error, we have cosidered two trasmsio oe rages, amely, 5.5 m ad 4.5results m. The agle magetometer deflectio due to evirometal oe cosidered asm±0 degrees. Figure 4a shows that as deflectio dtace betwee sesor odes ad sik amely, 5.5 ad 4.5 m. The agle magetometer due to evirometal oe cosidered as ±0 degrees. Figure 4a shows that as dtace betwee sesor odes ad sik odes icreasig due to multi-hop commuicatio ad error, but both s cosidered as ±0 degrees. Figure 4a shows that as so betwee sesor odes ad have sik odes icreasig due to multi-hop commuicatio ad sodtace error, but both s have same performace. Figure 4b shows that, due to limited wireless rage, DWL odes icreasig due to multi-hop commuicatio ad so error, but both s have same performace. Figure 4b shows that, due to limited wireless rage, DWL caot reach all etworkfigure odes ad,shows for cosidered parameters, three odes caot be localized. same performace. that, due to limited wireless rage, DWL caot reach all etwork odes4b ad, for cosidered parameters, three odes caot be localized. So, for se three odes localizatio error ifiite. Figure 4b shows that I-DWL So, for se three odes localizatio error ifiite. Figure 4b shows that I-DWL s ca reach se three odes ad localize m well. s ca reach se three odes ad localize m well.

13 Sesors 07, 7, caot reach all etwork odes ad, for cosidered parameters, three odes caot be localized. So, for se three odes localizatio error ifiite. Figure 4b shows that I-DWL s ca Sesors reach 07, se 7, 630 three odes ad localize m well. 4 6 Sesors 07, 7, Figure 3. Time ad eergy graph for full etwork localizatio. Figure 3. Time ad eergy graph for full etwork localizatio. Figure 4. Localizatio error. (a Without coectivity problem, (b With some coectivity problem Figure Localizatio error. (a (a Without coectivity problem, (b (b With some coectivity problem. 5. Coclusios Coclusios Coclusios The DWL uses two UWB radios with each ode ad requires a miimum two localized odes for localizatio a ulocalized ode, where all four radios two The The DWL DWL uses uses two two UWB UWB radios radios with with each each ode ode ad ad requires requires a miimum miimum two two commuicatig odes should be able to commuicate with remaiig three radios. We have localized localized odes odes for for localizatio localizatio a ulocalized a ulocalized ode, where ode, all four where radios all four two commuicatig radios two improved DWL by usig a magetometer o oe radios each ode, such that odes commuicatig should beodes able toshould commuicate be able to with commuicate remaiig with three remaiig radios. We three have radios. improved We have I-DWL requires oly oe ode to fully localize a ulocalized odes without strict DWL improved DWL by usig a magetometer by usig a magetometer o oe radios o oe each radios ode, such each that ode, such I-DWL that requiremet coectivity amog all four radios two commuicatig odes. Our results also I-DWL verify requires fact that olyrequires I-DWL oe odeoly outperforms to fully oe ode localize to fully i terms a ulocalized localize a time, eergy, odes ulocalized without odes umber strict without localized requiremet strict odes, requiremet localizatio coectivity error, amog coectivity amog all four radios two commuicatig odes. Our results also verify fact that ad I-DWL so all o. four I radios outperforms comparo with two i terms commuicatig DWL time,, odes. eergy, I-DWL Our results umber requires also localized almost verify odes, half fact localizatio umber that I-DWL error, semi-localizatios outperforms i terms ad so o. I comparo i a low-oe time, eergy, with eviromet, umber DWL, 60% localized I-DWL whole odes, etwork localizatio requires localizatio error, ad almost half time, so umber 70% o. I comparo semi-localizatios eergy, with ad ca DWL i localize, a low-oe more eviromet, umber I-DWL etwork requires almost 60% whole odes etwork with half a maitaied umber semi-localizatios localizatio localizatio time, 70% error i i eergy, cases a low-oe whe ad DWL eviromet, ca localize 60% more caot whole umber reach etwork all etwork etwork localizatio odes odes. time, 70% eergy, ad ca localize more umber etwork odes with a maitaied localizatio with error a maitaied i cases whe localizatio error i cases whe DWL caot reach all etwork odes. Author DWL Cotributios: Abdul caot Aziz reach developed all etwork th odes. idea, amesh Kumar helped i simulatio work ad Iwhee Joe guided ad supported well at every stage. Author Cotributios: Abdul Aziz developed th idea, amesh Kumar helped i i simulatio work ad Iwhee Clicts Joe guided Iterest: ad The supported authors declare well at at every o clict stage. iterest. Clicts Iterest: The The authors declare o o clict iterest. efereces. efereces Alhmiedat, T.A.; Yag, S.H. A Survey: Localizatio ad Trackig Mobile Targets through Wireless Sesors Network. I Proceedigs 8th Aual Postgraduate Symposium o Covergece. Alhmiedat, T.A.; Yag, S.H. A Survey: Localizatio ad Trackig Mobile Targets through Wireless Sesors Telecommuicatios, Networkig ad Broadcastig, Liverpool, UK, 8 9 Jue 007; pp. 6. Network. I Proceedigs 8th Aual Postgraduate Symposium o Covergece. Akyildiz, I.F.; Su, W.; Sakarasubramaiam, Y.; Cayirci, E. A Survey o Sesor Networks. IEEE Commu. Telecommuicatios, Networkig ad Broadcastig, Liverpool, UK, 8 9 Jue 007; pp. 6. Mag. 00, 40, Akyildiz, I.F.; Su, W.; Sakarasubramaiam, Y.; Cayirci, E. A Survey o Sesor Networks. IEEE Commu. Mag. 00, 40, 0 4.

14 Sesors 07, 7, efereces. Alhmiedat, T.A.; Yag, S.H. A Survey: Localizatio ad Trackig Mobile Targets through Wireless Sesors Network. I Proceedigs 8th Aual Postgraduate Symposium o Covergece Telecommuicatios, Networkig ad Broadcastig, Liverpool, UK, 8 9 Jue 007; pp. 6.. Akyildiz, I.F.; Su, W.; Sakarasubramaiam, Y.; Cayirci, E. A Survey o Sesor Networks. IEEE Commu. Mag. 00, 40, 0 4. [Crossef] 3. Chadrasekhar, V.; Seah, W.K.; Choo, Y.S.; Ee, H.V. Localizatio i Uderwater Sesor Networks: Survey ad Challeges. I Proceedigs st ACM Iteratioal Workshop o Uderwater Networks, Los Ageles, CA, USA, 5 September 006; pp Erol-Katarciz, M.; Mouftah, H.T.; Oktug, S. A survey architectures ad localizatio techiques for uderwater acoustic sesor etworks. IEEE Commu. Surv. Tutor. 0, 3, [Crossef] 5. Li, Z.; Li,.; Wei, Y.; Pei, T. Survey Localizatio Techiques i Wireless Sesor Networks. If. Techol. J. 00, 9, [Crossef] 6. Liu, H.; Darabi, H.; Baerjee, P.; Liu, J. Survey Wireless Idoor Positioig Techiques ad Systems. IEEE Tras. Syst. Ma Cyber. 007, 37, [Crossef] 7. Amudso, I.; Koustoubos, X.D.; Amudso, I.; Koustoubos, X.D. A Survey o Localizatio for Mobile Wireless Sesor Network. I Mobile Etity Localizatio ad Trackig i GPS-Less Eviromets; Spriger: Berli/Heidelberg, Germay, 009; pp Afzal, S. A eview Localizatio Techiques for Wireless Sesor Networks. J. Basic Appl. Sci. es. 0,, Garg, V.; Jhamb, M. A eview Wireless Sesor Network o Localizatio Techiques. It. J. Eg. Treds Techol. 03, 4, Samata, S.; Tembhare, P.U.; Pote, C.. A Survey o 3d Localizatio i Wireless Sesor Networks. It. J. Comput. Eg. es. 03, 3, Poudyal, L.; Se, B. A Survey o Localizatio ad Coverig Techiques i Wireless Sesor Networks. It. J. Comput. Appl. 03, 67, 3 7. [Crossef]. Patwari, N.; Ash, J.N.; Kyperoutas, S.; Hero, A.O.; Moses,.L.; Correal, N.S. Locatig odes. IEEE Sigal Process. Mag. 005,, [Crossef] 3. Doherty, L.; Pter, K.S.J.; El Ghaoui, L. Cove Positio Estimatio i Wireless Sesor Networks. IEEE INFOCOM 00, 3, Shag, Y.; Fromherz, M.P.J.; uml, W.; Zhag, Y. Localizatio from Mere Coectivity. I Proceedigs Iteratioal Symposium o Mobile Ad Hoc Networkig ad Computig (MobiHoc, Aapol, MD, USA, 3 Jue 003; pp Savvides, A.; Park, H.; Srivastava, M. The Bits ad Flops Nhop Multilateratio Primitive for Node Localizatio Problems. I Proceedigs st ACM iteratioal Workshop o Wireless Sesor Networks ad Applicatios (WSNA, Atlata, GA, USA, 8 September 00; pp.. 6. Zhog, Z.; Luo, D.-Y.; Liu, S.-Q.; Fa, X.-P.; Qu, Z.-H. A Adaptive Localizatio Approach for Wireless Sesor Networks Based o Gauss-Markov Mobility Model. Acta Autom. Si. 00, 36, [Crossef] 7. Gao, G.Q.; Lei, L. A Improved Node Localizatio Algorithm Based o DV-HOP i WSN. I Proceedigs IEEE Iteratioal Cerece o Advaced Computer Cotrol (ICACC, Sheyag, Chia, 7 9 May 00; pp Awad, A.; Fruzke, T.; Dressler, F. Adaptive Dtace Estimatio ad Localizatio i WSN Usig SSI Measures. I Proceedigs 0th Euromicro Cerece o Digital System Desig Architectures, Methods ad Tools (DSD 007, Lubeck, Germay, 9 3 August 007; pp Zhao, Q.; Yag, L. Two-step TOA estimatio method for UWB based wireless sesor etworks. J. If. Comput. Sci. 0, 9, og, P.; Sichitiu, M.L. Agle Arrival Localizatio for Wireless Sesor Networks. I Proceedigs 3rd Aual IEEE Commuatios Society o Sesor ad Ad Hoc Commuicatios ad Networks, esto, VA, USA, 8 Septermber She, Y.; Wi, M.Z. Fudametal limits widebad localizatio Part I: A geeral framework. IEEE Tras. If. Theory 00, 56, [Crossef]

15 Sesors 07, 7, Bartoletti, S.; Dai, W.; Coti, A.; Wi, M.Z. A mamatical model for widebad ragig. IEEE J. Sel. Top. Sigal Process. 05, 9, 6 8. [Crossef] 3. Dardari, D.; Coti, A.; Ferer, U.J.; Giorgetti, A.; Wi, M.Z. agig with ultrawide badwidth sigals i multipath eviromets. Proc. IEEE 009, 97, [Crossef] 4. Gezici, S.; Tia, Z.; Giaak, G.B.; Kobayashi, H.; Molch, A.F.; Poor, H.V.; Sahioglu, Z. Localizatio via ultra-widebad radios: A look at positioig aspects for future sesor etworks. IEEE Sigal Process. Mag. 005,, [Crossef] 5. Akca, H.; Evredilek, C. GPS-free directioal localizatio via dual wireless radios. It. J. Comput. Commu. 0, 35, [Crossef] 6. NS- Simulator Website. Available olie: (accessed o March Park, J.; Demaie, E.D.; Teller, S.J. Movig-Baselie Localizatio. I Proceedigs 7th Iteratioal Cerece o Iformatio Processig i Sesor Networks (IPSN 08, St. Lou, MO, USA, 4 April 008; pp Alavi, B.; Pahlava, K. Modelig TOA-based dtace measuremet error usig UWB idoor radio measuremets. IEEE Commu. Lett. 006, 0, [Crossef] 07 by authors. Licesee MDPI, Basel, Switzerlad. Th article a ope access article dtributed uder terms ad coditios Creative Commos Attributio (CC BY licese (